AI developers v creators and the Data (Use and Access) Act 2025

The Data (Use) and Access Bill received royal assent on 19 June 2025 and has become the Data (Use and Access) Act 2025. Although outside of the scope of the Bill as introduced, for much of May and June, debate around the Bill centred on the conflict between AI development and copyright law. The amendments tabled by the House of Lords sought to impose greater transparency over the use of data by AI companies in the development of their products. Baroness Kidron, a filmmaker, hoped to shift the legal landscape in favour of the creative industries, which she argued have, thus far, had no recourse against the unpermitted use of their materials by the AI sector.

AI and copyright

Some AI developers train their models on data obtained from the internet in a process known as scraping or data mining. Scraping is indiscriminate, and so the data used by AI developers will often contain large amounts of copyrighted material.

Currently, the rights that creators hold in their works are derived from the Copyright, Designs and Patents Act 1988 ("CDPA"). The CDPA has not been amended in response to the AI boom, no specific AI legislation has been introduced in the UK (unlike in the EU), and the UK courts have yet to rule on the use of copyrighted materials in AI training. Therefore, it is unclear whether AI businesses’ use of copyrighted materials infringes on the rights of creators.

AI companies and some experts have argued that their use of copyrighted works is permitted under UK law. Others have argued that the UK does not have jurisdiction to rule on such matters if the AI companies are (and the training of AI products takes place) outside the UK (e.g. in the US). However, this position may face increased scrutiny in light of the global reach of AI outputs and the potential accessibility of those outputs in the UK market. Those in the creative sector, however, take the position that such use is infringing their rights. Until recently, it seemed that Getty Images v Stability AI, the first copyright infringement case to be brought by a rightsholder against an AI developer in the UK, was poised to bring clarity to the law. The potential for the case to change the status quo was, however, diminished significantly when, last week, it was reported that Getty was no longer pursuing its direct copyright infringement claim in the High Court of Justice dispute.

One fact that those on all sides of this debate can agree on is that this uncertainty is harmful to both the creative industries and AI developers. Rightsholders feel that they have lost control over their works and desire to be compensated for their use in AI training (particularly where that use is commercial in nature), and AI developers have potentially been hesitant to fully invest in AI development in the UK for fear of future litigation.

The bill

Initially, the Data (Use and Access) Bill was not intended to address AI issues. The Bill, as introduced, focused on modernising the use of data across the UK in a multitude of ways.

Under the proposed amendment from the House of Lords, AI companies would have been required to publish information regarding the data used by companies in training their AI models and to respond to requests by copyright holders for such information. These changes were intended to create more transparency in the AI training process, enabling rightsholders to monitor the use of their works and assert their rights where appropriate. The House of Commons rejected the amendment, stating that it would wait until the government publishes a response to its recent AI and Copyright Consultation (please see our recent blog) before legislating on the issue.

The House of Lords thereafter proposed several amendments along the same lines, each of which were rejected by the House of Commons. The House of Lords ultimately withdrew their transparency amendments, and the Bill was allowed to pass to royal assent. However, while there was deadlock on the transparency issue, the House of Lords were able to push through an amendment which will require the Secretary of State to publish an assessment of the potential economic impacts of the policy options set out in the AI and Copyright Consultation. This requirement, while less prescriptive than the original proposals, keeps the issue on the legislative agenda and may influence future reform.

What's next?

The AI and Copyright Consultation set out several options to address the use of scraped data in AI training. The Government’s preferred option is to create a data mining exception in the CDPA in conjunction with measures on transparency. Under these measures, rightsholders would be able to more easily identify whether their work was being used in the training of AI models and would be able to prevent that use by reserving their rights (i.e. by opting out of their copyright protected works being capable of being used by AI companies). The Government believes that this approach would strike a reasonable balance between the interests of creatives and AI companies.

However, there has been further backlash from the creative sector, who believe that a default opt-in (and requirement to opt-out) is the wrong approach, and will prejudice less sophisticated creatives who might not be aware of, or understand, the opt-out process, or have the resources to spend on legal support to help them with the process. In this context, some stakeholders have called for a “robust rights-respecting AI framework” that would embed creator consent as a default rather than an afterthought. Whether the UK government is willing to adopt such a model remains to be seen. What is clear is that the question of how best to balance innovation with copyright protection remains one of the most pressing policy challenges in the digital economy.

For now, then, the status quo remains and both AI developers and the creative industries are left to navigate a murky and unsettled legal landscape – with all sides hopeful that clarity, in one form or another, is on the horizon.

If you are interested in exploring the key features of the Data (Use and Access) Act 2025, please see our separate blog. If you'd like any more information on AI and how it impacts your business, please contact a member of the IP Technology & Data team or check out our AI hub.